Industrial inspection

Ultrasonic thickness measurement: where to place CMLs and how to hold them

Where to place thickness monitoring points based on the damage mechanism, how many to set, and how to guarantee you re-measure the exact same spot.

7 min read

Turbines in the machine hall of a power station
Turbines in the machine hall of a power station

A thickness monitoring point (what the English-language literature calls a condition monitoring location) is a precise physical spot where a reading is taken at regular intervals. Its quality does not come from the accuracy of the instrument but from two things: is it in the right place, and can you find the exact same spot from one campaign to the next.

The essentials

A badly placed condition monitoring location measures an area that is not degrading: it reassures without informing. A badly marked one quietly shifts position over successive campaigns: it produces false trends that nothing flags. The second fault is more serious than the first, because it is invisible and because it contaminates the whole history. Location marking matters more than the number of points.

Where to place a CML: follow the mechanism, not the access

The first instinct is to put the points wherever you can reach: at chest height, on a straight run, next to a walkway. That is understandable, and it is the surest way to see nothing.

A useful point sits where degradation is expected, which assumes you know what you are looking for. The areas to watch differ according to the damage mechanism:

  • Erosion and erosion-corrosion: bends, tees, reducers, downstream of control valves, wherever the flow changes direction or cross-section, where turbulence attacks the material.
  • General corrosion: the low points of vessels, horizontal runs of pipework where water sits, condensation zones.
  • Under-deposit attack: low points, dead legs, low-velocity sections where deposits build up.
  • Corrosion under insulation: areas where the insulation is damaged or breached, under fixing clamps, at penetrations.
  • Fatigue: welds, fixing points, areas subject to vibration or thermal cycling.

This list makes one important point clear: a thickness monitoring location does not detect everything. A fatigue crack does not show up on an ultrasonic thickness reading. Choosing the technique according to the defect you are looking for is part of the job. That is the subject covered in risk-based inspection.

P1 P2 P3 Points recorded · straight runs, easy access Erosion: change of direction no monitoring point Eight years of stable readings. The points did not lie: the measurement plan was wrong.
Points set on accessible straight runs faithfully measure an area that is not degrading.

How many points to set

The question always comes back, and the honest answer is that there is no universal number. What does exist is a line of reasoning.

A single point on a piece of equipment gives you a value, not a condition. It lets you neither distinguish general degradation from a localised attack, nor tell whether the reading you have taken is representative. Two or three points on a homogeneous area start to give you a picture.

At the other extreme, multiplying points has a real cost: campaign time, surface preparation, data entry, and above all the follow-up burden. A plant with hundreds of points whose evolution nobody exploits is worth less than a plant with a few dozen points that are genuinely tracked.

The right sizing is reasoned through as follows:

SituationPoint density
Homogeneous area, general degradation expected, stable historyLow: a few representative points are enough
Complex geometry, turbulence, localised degradation expectedHigh: cover every configuration at risk
Equipment with high consequence of failureHigh, whatever the probability
Equipment with no historyHigh at first, then reduced once the behaviour is known
Long, stable history with a comfortable marginReduction possible, with a written justification

Table scrolls horizontally on small screens.

The last row deserves attention: reducing the number of points is a legitimate decision, provided it rests on data and is documented. Without a usable history, no reduction is defensible, neither in front of an auditor nor in front of your own management.

On the plant floor

What a badly placed point reveals

A pipe carrying a laden fluid has been monitored for eight years by three points on straight, easily accessible runs. The readings are remarkably stable: the loss of wall thickness is small and steady, and the margin is comfortable.

During a turnaround, an extended survey is carried out on the bends, which had never been measured before. The wall thickness there is markedly lower: erosion has concentrated where the fluid changes direction, exactly where nobody was looking.

The three original points did not lie. They faithfully measured an area that was not degrading. The fault was not in the measurement, it was in the measurement plan. Eight years of reassuring readings had reinforced the conviction that the equipment was fine.

Location marking: the condition without which nothing holds

A point exists only if it can be found again. That means three things, and all three have to travel with the inspection report.

A unique, stable identifier. It has to belong to the operator, not the contractor. If the reference changes when the contractor changes, the history breaks in two and nobody notices for several years.

A physical or documentary marking. Marking on the equipment, a dimensioned location sketch, a reference photograph. The sketch has to be attached to every campaign and reused as-is by whoever comes next.

The measurement conditions. Technique used, surface preparation, instrument. A change of technique introduces a systematic offset that will be read as degradation or as improvement if it is not documented.

  • Letting the contractor number the points.Their numbering belongs to them and changes with them: the reference has to be the operator's.
  • Placing points where access is easy,which means measuring where nothing is happening.
  • Keeping only the minimum thickness of the campaign.The minimum may come from a different point each time, and the comparison then means nothing: values have to be kept point by point.
  • Multiplying points without exploiting the readings.The workload rises, the knowledge does not.
  • Forgetting to record a change of technique.The offset introduced will be interpreted as a real change.
  • Reducing the number of points without a written justification.The decision may be sound, but it becomes indefensible in an audit if the reasoning has not been kept.

Keeping a measurement plan alive

A measurement plan is not fixed. It has to evolve with what you learn.

01

Start from the expected mechanism

Before placing a point, state what you are looking for. That one sentence, written once, guides everything else and explains why a given point exists.

02

Fix the references once and for all

Operator identifier, dimensioned sketch, photograph. It is the most thankless work in the file and the most profitable over ten years.

03

Reconcile each new campaign against the last one

An isolated value teaches you nothing. The reconciliation has to happen when the report arrives, not at an annual review that will never take place. See centralise inspection reports.

04

Treat anomalies as information

A wall thickness that rises beyond the measurement uncertainty signals a problem with the point, the technique or the data entry. It has to trigger a check, never a silent smoothing over.

05

Revise the plan when the knowledge changes

A defect found elsewhere, a change of fluid, a change of operating regime: each can justify adding or moving points. A measurement plan unchanged for fifteen years is rarely a good sign.

What a tool brings to this particular subject

Point-by-point monitoring is exactly the kind of task that spreadsheets handle badly. Every campaign arrives in a different format, the point-to-point correspondences are done from memory, and the person who keeps the file ends up being the only one who knows how it works.

Integrity software brings three things here: it ties each value to an identified point rather than to a spreadsheet row, it keeps the link back to the original report so that any anomaly can be arbitrated, and it applies the reconciliation across the whole plant rather than only to the equipment that is watched closely.

That is what a tool brings when it takes in the inspection reports you already have: the linking of values to identified points rather than to spreadsheet rows, and the retention of the original document to arbitrate any anomaly.

Sources and references

Condition monitoring locations: the body of Inspectioneering articles on the placement and optimisation of monitoring points, the reference in English on the subject. CML resources

API 570: the in-service piping inspection code, which frames the building of a monitoring-point plan and the calculation of the associated corrosion rates.

Should the points be physically marked on the equipment?

That is the most robust solution when it is possible. On insulated equipment or in an aggressive environment the marking wears off: you then have to rely on a dimensioned sketch and a reference photograph, attached to every campaign.

Can points be added during monitoring?

Yes, and it is even desirable when the knowledge evolves. An added point starts its own history: you must not merge it with a neighbouring point on the grounds that it sits close by. Two distinct points remain two distinct series.

What do you do when two contractors have numbered differently?

Draw up a correspondence table, have it validated by someone who knows the equipment, and keep it. That table is part of the file just as much as the reports: without it, nobody will be able to reconstruct the reasoning in five years' time.

How many points for a new piece of equipment?

More than you will need in the end. On equipment whose behaviour you do not yet know, broad coverage of the areas at risk lets you identify where the degradation actually concentrates. The plan can then be tightened, with justification.

Written by Adama CamaraAI Consultant · Industry · view profile

Published on June 23, 2026

Support

Custom AI systems for industry

Agents that put your data to work and extend your existing tools. Designed and run on site, off the network.

Visit Assets 4.0